ZIP8 loss impairs macrophage-mediated phagolysosomal removal of bacteria and is overcome by butyrate supplementation

Abstract

Zinc deficiency impacts billions of people and contributes significantly to the increased incidence of community-acquired pneumonia worldwide. Myeloid cells require the zinc transporter ZIP8 for proper host defense. Previously, we observed that infection with S.pneumoniae in myeloid-specific Zip8 knockout mice (Zip8KO) results in increased bacterial burden and mortality despite increased recruitment of macrophages into the lung. Here, we reveal that the lungs of infected Zip8KO mice generate a unique population of dysfunctional macrophages with defects in phagolysosomal function and cell survival. In particular, Zip8KO bone marrow-derived macrophages have increased bacterial accumulation due to deficits in lysosomal number and function via defective mTORC1/TFEB signaling. Knowing that labile Zn cannot enter the cytosol through ZIP8 and that ZIP8 loss impairs butanoate synthesis by the gut microbiome, both previously reported by our group, we reveal an alternative treatment strategy via extended oral phenylbutyrate supplementation. Despite ongoing ZIP8-mediated impairment of lung host defense, phenylbutyrate restored macrophage-mediated bacterial clearance and improved host outcomes. Given the high incidence of diet-induced Zn deficiency and the rs13107325 ZIP8 defective variant allele in humans, future investigations that foster preventive, patient-centered treatment strategies that counter immune dysfunction due to Zn dyshomeostasis are warranted.

ZIP8 loss decreases intracellular zinc uptake in macrophages in response to bacterial pneumonia and impairs phagolysosomal function and bacterial clearance. Despite ongoing zinc dyshomeostasis, prolonged butyrate supplementation corrects bacterial clearance and improves survival.

Introduction

A primary cause of morbidity and mortality worldwide is bacterial community-acquired pneumonia (CAP). The most common bacterial cause of CAP in the U.S. is Streptococcus pneumoniae, which results in increased hospitalizations and mortality^1–4. Compromised immune function, frequently associated with aging and comorbidity, is a leading cause of CAP⁵. Zinc (Zn) is an essential trace element and is required to maintain proper immune function. Populations that are most vulnerable to CAP also exhibit a high incidence of Zn deficiency^6–8, which is estimated to impact over 2 billion people worldwide, including approximately 20–30% of the U.S. population⁹. Consistent with this, others have reported that dietary Zn deficiency increases susceptibility to gastrointestinal tract infections¹⁰ and pneumonia^11,12, whereas the incidence of pneumonia and other infections is decreased with Zn supplementation^13–15.

Zn homeostasis in mammals is regulated by two major solute-linked carrier (SLC) Zn transporter gene families comprised of 14 SLC39 genes, also known as ZIP transporters, and 10 SLC30 genes, also known as ZnT transporters. SLC30 transporters decrease cytosolic Zn levels, whereas SLC39 transporters promote extracellular uptake or intracellular release from organelles, thereby increasing cytosolic Zn content¹³. ZIPs have collectively been shown to regulate the activity of enzymes, receptors, transcription factors, and cytokine- and growth factor-mediated signaling pathways via Zn mobilization into the cytosol^14,15. ZIP8 is the only ZIP induced upon pathogen detection and is required by myeloid and parenchymal cells to defend against pathogens through mobilization of Zn in addition to iron and manganese^16–19. This is of clinical relevance because recent genome-wide association (GWAS) studies have revealed that a frequently occurring defective ZIP8 variant allele (rs13107325; Ala391Thr risk allele) is highly associated with inflammation-based immune disorders^20,21 and bacterial infections²². However, the mechanisms by which ZIP8 and Zn influence host defense and pathogen removal remain unclear. Consistent with this, we revealed that myeloid-specific Zip8 knockout (Zip8KO) mice have a significant increase in morbidity and mortality in response to pneumococcal pneumonia and that despite increased recruitment of macrophages to the lung, tissue bacterial burden increased and disseminated to other vital organs^23,24.

Macrophages are the main phagocytic cells of the innate immune system and destroy pathogens through the phagolysosomal pathway²⁵. Lysosomes serve as a central signaling hub that controls molecular decision-making through mammalian target of rapamycin complex 1 (mTORC1). The transcription factor EB (TFEB) is the master transcriptional regulator of lysosome biogenesis and is controlled upstream by mTORC1. Upon dephosphorylation, TFEB translocates to the nucleus, inducing transcription of phagolysosomal proteins required for lysosome formation and function^26–29. In particular, exposure to pathogens increases the expression and activation of TFEB in macrophages, leading to lysosome-mediated degradation of bacteria^30–33. However, the impact of Zn on macrophage lysosomal function via mTORC1/TFEB and bacterial eradication remains unknown. In a previous study, we discovered that the gut microbiome of Zip8KO mice is distinctly altered, leading to a significant reduction in the production of the short-chain fatty acid³⁴ (SCFA) butyric acid (BA)³⁵. Related to this, Schulthess and colleagues revealed that differentiation of macrophages in the presence of BA increased antimicrobial activity in part through alteration of mTORC1-dependent production of antimicrobial peptides³⁶. Based on these interrelated studies, we determined whether ZIP8 loss adversely affects mTORC1/TFEB-mediated signaling, bacterial eradication, and the host response to pneumonia and if so, whether BA supplementation restores macrophage-mediated host defense in the setting of Zn dyshomeostasis due to defective ZIP8 function.

In this investigation, we first reveal that a unique cluster of defective proinflammatory macrophages emerges in the lungs of infected Zip8KO mice that are predicted to harbor significant deficits in phagosome formation and promote cell death. Strikingly, infected Zip8KO BMDMs exhibited defective mTORC1/TFEB signaling and a corresponding reduction in lysosome numbers and enzymatic activity, culminating in decreased bacterial clearance. Based on this and knowing that countermeasures with Zn supplementation are compromised due to defective intracellular Zn uptake, we reveal for the first time that prolonged BA supplementation counters ZIP8-mediated defects and restores macrophage function and bacterial clearance in vitro and in vivo despite ongoing defects in Zn mobilization. Overall, this demonstrates that ZIP8 is essential for macrophage-mediated removal of bacteria and resolution of infection and reveals that innovative microbiome-derived strategies may foster the development of countermeasures to improve the host response against respiratory infections in the setting of Zn dyshomeostasis. Given the high incidence of dietary Zn deficiency and high frequency of the A391T ZIP8 variant allele, future studies are warranted to identify preventative treatment strategies that lower the overall incidence of CAP in vulnerable populations.

Results

Alteration of the lung macrophage landscape in the setting of ZIP8 loss

Knowing that Zip8KO mice have increased lung macrophages with reduced phagocytic capacity²³, we reasoned that ZIP8 loss may affect the collective composition and/or function of macrophage subpopulations. To characterize macrophage populations in the lung, we conducted single-cell RNA sequencing of perfused lung tissues isolated from uninfected WT and Zip8KO mice and infected WT and Zip8KO mice 72 h after intratracheal installation of 4 × 10⁸ CFU S. pneumoniae. Prior to sequencing, bronchoalveolar lavage fluid (BALF) was analyzed to confirm replication of previously published findings demonstrating increased myeloid cell counts in Zip8KO-infected mice compared to WT-infected mice²³. Using unbiased cluster analysis, we identified 24 distinct clusters (Fig. 1a). We then identified and characterized macrophage-specific clusters based on the expression of multiple proinflammatory (M1-like) (CD80, CD86, CD64, CD32, iNOS, CD40, CD83, MHCII, TLR4, IL1-R1, SOC3, TLR2) and anti-inflammatory (M2-like) (CD163, CD206, Arg1, Fizz1, Ym1, CD115, PPARg, CD301, PDL2) markers in addition to corresponding transcriptomic scoring profiles which revealed that clusters 1, 3, 4, 5, 6, 9, 10, 15, 17, 18, 19, and 21 were predominantly populated by macrophages of a proinflammatory phenotype (Fig. 1b and Supp. Table 1). Strikingly, cluster 10 primarily appeared only in the lungs of infected Zip8KO mice when compared to the other groups (Fig. 1c). Further examination of cluster 10 using an array of monocyte-derived macrophage (Adgre1, Itgam, Fcgr1) and tissue-resident macrophage (Adgre1, CD68, Itgam, Fcgr1, Itgax) markers revealed that it consists of a mixture of recruited and resident macrophages (Fig. 1d). Although the focus of the current study is on macrophage phenotype and function, identification of other immune cell populations was also conducted across the 24 clusters (Supplementary Fig. 1).

Fig. 1. Alteration of the lung macrophage landscape in the setting of ZIP8 loss.

Single-cell RNA sequencing analysis of perfused lung tissue from naïve and infected WT and Zip8KO mice (n = 1 per group). a Unbiased cluster analysis revealed 24 clusters across the four treatment groups. b Analysis of CD86 and CD163 expression revealed that clusters 1, 3, 4, 5, 6, 10, 12, and 17 are populated primarily by macrophages of an M1 phenotype. c Cluster 10 is observed primarily only in the Zip8KO-infected lung. d Violin Plot of Macrophage markers Adgre1 (F480), Itgam (Cd11b), Itgax (Cd11c), CD68, Fcgr1 (CD64), and SiglecF in Cluster 10 of the infected Zip8KO mouse lung demonstrates a mixed population of resident and recruited cells. e, f IPA analysis of clusters 1, 3, 4, 5, and 10 reveals that all clusters are defective in phagosome formation and that many pathways associated with M1 characteristics and cell death are prevalent in and unique to cluster 10. g–l FACS analysis of lung tissue identifying total leukocyte, neutrophils, and other myeloid-specific cell subtypes. Also see Supplementary Data 1, Fig. S1, and Table S1.

We next utilized Ingenuity Pathway Analysis (IPA) to gain insight regarding the functionality of each macrophage cluster. Strikingly, cluster 10 and the majority of other clusters exhibited a significant decrease in pathways associated with phagosome formation in response to infection (Fig. 1e and Table S2) and altered mTORC1 signaling (Fig. 1f), consistent with previously published in vivo studies demonstrating an increase in bacterial burden in the lung despite increased macrophage numbers²³. Also, consistent with past studies showing increased numbers of apoptotic cells in the lung of infected Zip8KO mice compared to infected WT counterparts²³, we revealed that cluster 10, in addition to other clusters, exhibited an increase in multiple networks associated with programmed cell death (Fig. 1e and Table S2). We also observed that cluster 10 had a significant increase in GM-CSF and interferon-signaling networks, relative to most other macrophage clusters, consistent with an activated, proinflammatory phenotype.

Based on macrophage/monocyte-specific mRNA profiles, we next isolated lung tissue cells and analyzed samples using flow analytical cell sorting to determine the type and abundance of cell populations between treatment groups (Fig. S2 and S3). As anticipated, both infected WT and Zip8KO mice had a robust myeloid response to infection. Compared to infected WT mice, Zip8KO counterparts had a significant increase in overall myeloid cells, including neutrophils, with modest increases in other myeloid populations, but did not achieve statistical significance. (Fig. 1g–l and Table S3). Overall, these data reveal that although there exist some differences in myeloid populations, the differences are not profound, unlike findings from scRNAseq and IPA analysis that revealed significant differences in the functionality of macrophage-populated clusters.

ZIP8 loss increases macrophage bacterial burden and cell death

Based on single-cell RNA sequencing and IPA analysis, we next enumerated S. pneumoniae, macrophages, and apoptotic macrophages in situ. WT and Zip8KO mice were administered 4 × 10⁸ CFU S. pneumoniae or phosphate-buffered saline (PBS) as a control and euthanized 48 h post-infection. Zip8KO mice compared to WT mice had more; F4/80+ macrophages (Fig. 2a lower left and right panels), bacteria (Fig. 2b, c), and TUNEL+/F480+ macrophages in the lungs following infection (Fig. 2d). Confocal microscopy also revealed that while many of the TUNEL+ cells were macrophages, there was also an increase in the number of TUNEL+ cells throughout the lung parenchyma of Zip8KO mice, presumably constituting apoptotic epithelial and endothelial cells. Consistent with this, Zip8KO mice exhibited a significant increase in circulating levels of surfactant protein D (SP-D) in the blood compared to infected WT counterparts, indicating increased barrier permeability in the lung (Fig. 2e), as well as an increase in morbidity based on their bacterial pneumonia assessment score (Fig. 2f and Table S4). Lung tissue Zn content was decreased at baseline in Zip8KO mice and moderately increased following infection, but without significance, and not to the same extent as infected WT mice (Fig. 2g).

Fig. 2. ZIP8 loss increases macrophage bacterial burden and cell death.

a Immunostaining of F480 (green), nuclear staining with DAPI (blue), TUNEL staining (red), and GFP-tagged bacteria (white) in naïve and infected (48 h post-infection) WT and Zip8KO mice lung sections at 63X (n = 4). Scale bars represent 100 µm. White arrows highlight TUNEL+ macrophages. b Mean fluorescence intensity of GFP-tagged D39 S. pneumoniae is significantly increased in Zip8KO mice (n = 4) compared to WT infected mice. c Zip8KO lungs have higher intracellular bacterial burden 18 h after infection as measured by CFU counts. d Mean fluorescence intensity of TUNEL+ macrophages is significantly increased in infected Zip8KO mice compared to infected WT mice (n = 4). e Serum SP-D levels are higher in infected Zip8KO mice compared to infected WT mice (n = 7–12). f Comparison of modified mouse clinical assessment scores at baseline, 24, and 48 h after infection. g Comparison of lung tissue Zn levels before and 48 h after infection. Data are represented as mean ± SEM for (b–d). Statistical analysis was performed using an unpaired t-test. (^*p < 0.02; ^**p < 0.006; ^****p < 0.0001). See Supplementary Data 1 and Table S4.

Zip8KO macrophages have reduced lysosomes, enzymatic activity, and bacterial clearance

Knowing that Zip8KO mice have altered lung macrophage composition upon infection and increased bacterial burden, we next focused on in vitro studies. To determine how macrophage function is regulated by ZIP8, we first isolated bone marrow from WT and Zip8KO mice and differentiated monocytes into macrophages (BMDMs) in the presence of GM-CSF to model the lung environment observed in vivo in infected Zip8KO mice. We then exposed BMDMs to S. pneumoniae for 60 min at an MOI of 10:1 and evaluated lysosome numbers, lysosomal enzyme activity, and intracellular bacterial burden. Strikingly, Zip8KO BMDMs had a significant reduction in lysosome numbers based on LAMP-1 immunostaining at baseline and in response to infection compared to WT BMDMs (Fig. 3a, b). Following extended incubation out to 18 h, Zip8KO BMDMs also had a significant increase in intracellular bacteria (Fig. 3c). We then compared lysosomal enzymatic activity following infection. We observed that Zip8KO BMDMs exhibited a marked reduction in enzymatic activity in response to infection (Fig. 3d, e) despite no significant differences in mature cathepsin B or L expression at baseline or at 18 h post-infection (Fig. 3f, g). Most striking, and consistent with the general reduction observed in lysosomal enzymatic activity, both cathepsin B and L enzymatic activity were significantly decreased in Zip8KO BMDMs compared to WT BMDMs at 18 h post-infection (Fig. 3h, i), which also corresponded with a reduction in LysoTracker staining within 15 min following infection (Fig. 3j, k). Collectively, these data suggest that decreased bacterial clearance is caused by a reduction in lysosome numbers with defective enzymatic function, perhaps due to insufficient acidification.

Fig. 3. Zip8KO macrophages have reduced lysosomes, enzymatic activity, and bacterial clearance.

a Immunofluorescence imaging of LAMP-1 (red) at 15, 30, and 60 min post-infection. The nucleus is stained with DAPI. Scale bars represent 100 µm. b Quantification of LAMP-1 mean fluorescence intensity. c Zip8KO BMDMs have higher intracellular bacterial burden 18 h after infection as measured by CFU counts. d DQ-BSA (red) stained BMDMs, non-infected and 18 h post-infection. The nucleus is stained with DAPI. Scale bars represent 50 µm. e Quantification of lysosomal enzymatic activity in Zip8KO BMDMs is decreased compared to WT 18 h post-infection. f Immunoblot of mature cathepsin B and L, and β-actin proteins from non-infected and 18 h post-infection. g Densitometric analysis of mature cathepsin B and L standardized to Β-actin from noninfected and 18 h after infection. h, i Quantification of cathepsin B (h) and L (i) activity. Zip8KO BMDMs have decreased cathepsin B & L activity at baseline and 18 h post-infection. j Immunofluorescence imaging of lysosomes using lysotracker (red) at 15- and 30-min post-infection of D39. The nucleus is stained with DAPI. Scale bars represent 50 µm. k Mean fluorescence intensity of lysotracker staining is decreased in Zip8KO BMDMs at 15- and 30-min post-infection; however, at 60 min there is no difference compared to WT at the same time points. Data is representative of 2–3 independent experiments and are represented as mean ± SEM. Statistical analysis was performed using an unpaired t-test for graphs (b, c, e, g, h). (^*p < 0.05, ^**p < 0.009, ^****p < 0.0001).

Zip8 loss in macrophages reduces intracellular Zn content and alters mTORC1 signaling

Knowing that lysosomal numbers and enzymatic activity are reduced, we next determined whether ZIP8, or lack thereof, has a direct impact on mTORC1/TFEB signaling, the master transcriptional regulator for lysosomal biogenesis, which was also identified previously by IPA analysis (Fig. 1f). As predicted, we observed a significant increase in phospho-mTORC1 expression in Zip8KO BMDMs compared to WT BMDMs at baseline and following infection (Fig. 4a, b). Consistent with this, in comparison to WT BMDMs, TFEB nuclear localization was decreased in noninfected and infected Zip8KO BMDMs as determined by flow cytometry (Fig. 4c, d), as well as Western blotting of nuclear fractions (Fig. 4e, f). To pharmacologically validate these findings, pretreatment with the mTORC1 inhibitor rapamycin (250 nM for 24 h) significantly increased bacterial clearance in Zip8KO BMDMs when compared to non-rapamycin-treated Zip8KO BMDMs (Fig. 4g).

Fig. 4. ZIP8 loss in macrophages reduces intracellular Zn influx and alters pmTOR/TFEB signaling.

a, b Immunoblot (a) and quantification (b) of mTOR shows Zip8KO BMDMs have increased phosphorylation of mTOR compared to WT BMDMs at 1 h post-infection. c Cytoplasmic and nuclear staining of total TFEB in WT BMDMs from FACs analysis. Images include brightfield, TFEB-red, DAPI-stained nucleus, and a merge of TFEB/DAPI. d Zip8KO BMDMs have less nuclear total TFEB 1 h post-infection. e, f Western analysis similarly reveals a decrease in nuclear TFEB Zip8KO BMDMs before and after infection. g Pre-treatment with mTOR inhibitor, 250 nM rapamycin for 24 h, reduced bacterial burden and restored clearance in Zip8KO BMDMs to a level similar to WT BMDMs. h Intracellular Zn concentration is lower in infected Zip8KO BMDMs compared to infected WT BMDMs. i Pretreatment of WT BMDMs with the Zn chelator TPA (1 µM) for 30 min increased bacterial burden following infection and also further increased bacterial burden in Zip8KO BMDMs. Data is representative of a minimum of two independent experiments and are represented as mean ± SEM. Statistical analysis was performed by an unpaired t-test for graphs (b, d, e–g). (^*p < 0.05, ^**p < 0.007, ^***p < 0.0006). Also see Supplementary Data 1 and Fig. 4a, b.

To determine the extent to which ZIP8 loss decreases macrophage cytosolic Zn, we measured intracellular Zn content under similar conditions at baseline and up to 6 h after BMDMs were exposed to S. pneumoniae. Under baseline conditions, there were no significant differences between intracellular Zn content in WT and Zip8KO BMDMs. However, immediately following the 60-min exposure and out to at least 6 h, intracellular Zn content significantly increased in WT BMDMs, but not to the same extent in Zip8KO BMDMs (Fig. 4h). To further validate that decreased intracellular Zn adversely impacts bacterial clearance, WT BMDMs were pretreated with the intracellular Zn chelating agent Tris(2-pyridylmethyl)amine (TPA)(1 µM) for 30 min before infection. Sequestration of intracellular Zn in WT BMDMs also resulted in a significant increase in intracellular bacteria when compared to non-TPA-treated WT cells, which was further increased in Zip8KO BMDMs (Fig. 4i). Knowing that ZIP8 also transports iron (Fe) and manganese (Mn), we measured the intracellular content for both and found that immediately following bacterial exposure, Zip8KO BMDMs had reduced intracellular Fe and Mn (Fig. S4a, b) although by 6 h the intracellular levels of each were identical to infected WT cultures, unlike Zn. Collectively, these findings demonstrate that decreased intracellular Zn content, either by defective ZIP8-mediated transport or intracellular chemical sequestration, causes defects in the ability of macrophages to clear bacteria and that this is due at least in part to defective mTORC1/TFEB signaling. Although deficits in intracellular Zn clearly contribute to defective bacterial clearance, we cannot rule out that other ZIP8 substrates (Fe and Mn) may also contribute to this effect.

Phenylbutyrate (PBA) supplementation restores proper host defense, bacterial clearance, and clinical outcomes in Zip8KO mice in response to pneumococcal pneumonia

ZIP8 is the only ZIP induced upon bacterial recognition and is required for proper innate immune defense against bacteria in myeloid cells and lung tissue^17–19,23. We have also previously found that preventive dietary Zn supplementation strategies with conventional inorganic zinc forms (ZnCl₂ and ZnSO₄) are ineffective, most likely because Zn cannot gain intracellular access through ZIP8 to bolster host defense. Knowing that butyrate enhances the anti-microbial properties of macrophages³⁶, and that gut microbiome-derived butyrate levels are reduced in Zip8KO mice³⁷, we compared the lung metabolomic profile of macrophage-specific clusters 1, 3, 4, 5, and 10 in uninfected and infected WT and Zip8KO mice. To accomplish this, we first developed predictive genome-scale metabolic (GSM) models of macrophage clusters (1, 3, 4, 5, and 10) using optimization techniques that included flux balance analysis (FBA)³⁸, flux variability analysis (FVA)³⁹, and flux sum analysis (FSA)⁴⁰ to predict the overall impact of butyrate, or lack thereof, as well as the impact of adding butyrate back into the models (Fig. 5). Under pre-infection conditions the Zip8KO (KOPBS) model exhibited a butyrate deficit, with reduced levels of metabolites including 2-hydroxybutyrate, 2-oxobutyrate, and (R)-3-hydroxybutanoate. At baseline, these changes were marginal, suggesting minimal differences exist in WT (WTPBS) and Zip8KO macrophages before infection (Fig. 5a, c). However, following infection (WT72 and KO72), significant reductions in butyrate levels were predicted, particularly in clusters 4, 5, and 10 (Fig. 5d, e). Notably, multiple butyrate-related metabolites decreased in Zip8KO models post-infection, although some metabolites, including crotonyl-CoA and 4-aminobutyrate, increased in clusters 1, 5, and 10 (Fig. 5d), highlighting the critical role of ZIP8-mediated Zn homeostasis in maintaining gut metabolic balance and its potential interrelationship with lung metabolic processes. Given that the Zip8KO models implicated butyrate deficiency both before and after infection, we investigated the effects of reintroducing butyrate. Butyrate supplementation significantly impacted reaction fluxes in both pre- and post-infection conditions (Fig. 5b, e, respectively). Notably, the reactions with substantial changes in fluxes were primarily associated with central carbon metabolism, including glycolysis, the pentose phosphate pathway, oxidative phosphorylation, and the tricarboxylic acid cycle. Additionally, butyrate supplementation to the Zip8KO models shifted the reaction fluxes closer to the WT levels (Fig. S5). Collectively, our computational models predict that butyrate supplementation has the potential to restore key metabolic pathways, thereby improving the physiological state of Zip8KO models and mitigating the metabolic dysregulation caused by ZIP8 loss and intracellular Zn deficiency in lung macrophages.

Fig. 5. In silico analysis identifies butanoate metabolism as a critical factor in Zn-mediated alteration of the host response to infection.

a Change in metabolite pool sizes across macrophage clusters in WT and Zip8KO before infection as predicted by metabolic modeling. Statistical analysis was performed using an unpaired t-test for graphs (a, d) (^**p < 0.05). b Change in flux space of Zip8KO models after adding butyrate before and after infection. Significant changes in flux differences before and after butyrate addition are denoted as mean flux difference ± standard deviation in (b, e). c Average change in metabolite pool sizes of butanoate metabolite intermediates of WT and Zip8KO macrophages before infection. Statistical significance is based on Fold Change, where (^*represents FC > = 1, ^**represents FC > = 3, and ^***represents FC > = 5). d Change in metabolite pool sizes across macrophage clusters in WT and Zip8KO post-infection as predicted by the metabolic models. Statistical analysis was performed using an unpaired t-test for graphs (a, d) (^**p < 0.05). e Change in flux space of Zip8KO models after adding butyrate in both post-infection conditions. Significant changes in flux differences before and after butyrate addition are denoted as mean flux difference ± standard deviation. f Average change in metabolite pool sizes of butanoate metabolite intermediates of WT and Zip8KO macrophages after infection. Statistical significance is based on fold change (^*represents FC = 1, ^**represents FC > = 3, and ^***represents FC > = 5).

Based on this, we evaluated whether PBA supplementation improves outcomes in Zip8KO mice in response to pneumococcal infection. Early attempts utilizing a one-time PBA intraperitoneal injection or administration of PBA via gavage for one week failed to improve outcomes in infected Zip8KO mice, resulting in identical outcomes as previously shown in Fig. 2. Based on this, we implemented an extended treatment period. WT and Zip8KO mice were administered PBA (30 mM) in their water ad libitum for 4 weeks. Treatment groups that received PBA were indistinguishable from vehicle-treated mice in terms of appearance, composition, and water intake. PBA-treated mice also received a one-time intraperitoneal injection of either PBA (400 mg/kg/100 µL) or the same volume of PBS at 24 h post-infection, since water intake significantly declines following infection. At baseline without infection, there were no differences in bronchoalveolar lavage (BAL) total and differential cell counts, BAL protein content, and pro-inflammatory cytokine levels between all treatment groups. Using a modified clinical assessment panel (Table S3), we observed at 24 and 48 h following infection that PBA-treated Zip8KO mice had a significant improvement (lower scores) in the majority of clinical parameters compared to vehicle-treated Zip8KO mice, while, as expected, there was no difference between the PBA-treated and untreated WT mice (Fig. 6a). As observed previously, vehicle-treated, infected Zip8KO mice exhibited a higher number of BAL cells, macrophages, and neutrophils compared to WT infected mice (Fig. 6b–d). Following prolonged treatment with PBA, Zip8KO mice had significantly decreased total BAL cell infiltrates, including a significant reduction in neutrophils compared to vehicle-treated Zip8KO mice at 48 h post-infection (Fig. 6d). A similar trend was observed with BAL macrophage counts (Fig. 6c), BAL protein (Fig. 6e), and CXCL1 (Fig. 6f), although these differences did not achieve statistical significance. Analysis of perfused lung tissue also revealed a significant decrease in total leukocytes and neutrophils in PBA-treated Zip8KO mice, but with unremarkable differences in monocyte/macrophage subpopulations as measured by flow cytometry (Fig. 6g, l and Table S5).

Fig. 6. Butyrate supplementation improves host defense in Zip8KO mice against pneumococcal pneumonia.

a PBA-treated and untreated WT and Zip8KO mice were infected with 4 × 10⁸ CFU S. pneumoniae, and the pneumonia clinical assessment score was assessed up to 48 h post-infection. b–f PBA-treated and untreated WT and Zip8KO mice were infected with 4 × 10⁸ CFU S. pneumoniae, and lungs were harvested and examined at 48 h post-infection and at baseline for total BAL cell count (b), total macrophage count (c), total neutrophil count (d), BAL protein (e), and CXCL1 levels (f). (n = 10-14 per group). g–l PBA-treated and untreated WT and Zip8KO mice were infected with 4 × 10⁸ CFU S. pneumoniae, and lungs were harvested and examined by FACS at 48 h post-infection and at baseline for total leukocytes, neutrophils, and other myeloid-specific subsets. Data are represented as mean ± SEM. Statistical analysis was performed by an unpaired t-test for graphs (a–f). (^*p < 0.05, ^**p < 0.004, ^***p < 0.008).

Consistent with these findings, microscopic analysis of lung tissue sections revealed a significant reduction in the number of F480+ macrophages, as well as TUNEL+/F480+ macrophages in PBA-treated Zip8KO mice when compared to vehicle-treated counterparts (Fig. 7a; UR panel vs LR panel; Fig. 7d) which was also observed in PBA-treated infected WT mice although to a lesser extent (Fig. 7a; UL panel vs LL panel; Fig. 7d). In addition, bacterial burden was significantly reduced in the lungs of Zip8KO mice following PBA administration (Fig. 7b, c). Compared to vehicle-treated Zip8KO mice, PBA-treated Zip8KO mice also exhibited a significant decrease in pulmonary permeability as shown by lower serum levels of SP-D (Fig. 7e), a known biomarker of lung damage⁴¹. PBA did not change lung tissue Zn levels in response to infection in either WT or Zip8KO mice (Fig. 7f).

Fig. 7. Prolonged PBA supplementation reduces bacterial burden and cell death in Zip8KO mice and improves outcomes.

a Immunostaining of F480 (green), nuclear staining with DAPI (blue), TUNEL (red), and D39 (white) staining in PBS or PBA-treated WT infected mice (upper left and lower left panels) and PBS or PBA-treated infected Zip8KO mice at 48 h following infection. Lung sections magnified at 63× (n = 10). Scale bars represent 100 µm. b Quantification of imaging of WT and Zip8KO mouse lungs at 48 h post-infection revealed a significant decrease in S. pneumoniae (D39) in Zip8KO mouse lungs following PBA treatment compared to vehicle control counterparts. c Similarly, quantification of Colony Forming Unit counts obtained from WT and Zip8KO mouse lungs at 48 h post-infection revealed a significant decrease in S. pneumoniae (D39) in Zip8KO mouse lungs following PBA treatment compared to vehicle control counterparts. d Quantification of apoptotic macrophages following PBA treatment significantly decreased in TUNEL-positive macrophages in infected WT and even more so in infected Zip8KO mice (n = 4). e Serum SP-D levels decreased in PBA-treated Zip8KO post-infection. (n = 9-12). f PBA had no effect on zinc levels in lung tissue post-infection. (n = 9-12). Data are represented as mean ± SEM. Statistical analysis was performed by an unpaired t-test. (^*p < 0.05, ^**p < 0.006 for graphs (b–e).

PBA exposure during differentiation of Zip8KO macrophages restores antibacterial defense

To determine the anti-microbial properties of PBA on macrophage-mediated clearance of bacteria, we examined its impact on WT and Zip8KO BMDMs relative to bacterial burden, mTORC1, TFEB nuclear localization, and lysosomal enzymatic activity. Consistent with previous reports³⁶ PBA exposure post-differentiation had no beneficial impact on the clearance of intracellular S. pneumoniae in WT and Zip8KO BMDMs (Fig. 8a); however, exposure of Zip8KO BMDMs to 100 µM PBA during differentiation for 7 days restored the ability of Zip8KO BMDMs to eradicate intracellular bacteria compared to untreated Zip8KO and PBA treated WT BMDMs (Fig. 8b). In a similar manner, PBA exposure during differentiation of Zip8KO BMDMs resulted in an mTORC1 phosphorylation pattern similar to WT cultures following infection (Fig. 8c, d) with a corresponding increase in TFEB nuclear localization (Fig. 8e) when compared to nonPBA-treated infected Zip8KO BMDMs although the extent of mTORC1 phosphorylation was not different between infected PBA treated WT and Zip8KO cultures (Fig. 8d). From a functional perspective, lysosomal enzymatic activity was significantly increased (Fig. 8f, g) including a significant increase in cathepsin L activity (Fig. 8h) in PBA-differentiated Zip8KO BMDMs compared to WT cells likely contributing to restoration of bacterial clearance in Zip8KO BMDMs.

Fig. 8. Supplementation with PBA during differentiation restores macrophage-mediated bacterial clearance.

a Treatment with PBA post-differentiation does not improve bacterial clearance in Zip8KO BMDMs. b PBA supplementation during differentiation improves bacterial clearance in Zip8KO BMDMs. c, d PBA normalizes the phosphorylation of mTORC1 in Zip8KO BMDMS, similar to that of WT BMDM levels following infection. e PBA treatment increases TFEB nuclear localization in Zip8KO BMDMs following infection. f Immunofluorescence images of DQ-BSA-stained WT and Zip8KO BMDMs with and without PBA supplementation before and after infection. g PBA treatment increases lysosome enzyme activity in Zip8KO BMDMs before and after infection. h PBA treatment increases Cathepsin L activity in Zip8KO BMDMs following infection. The data represented are three independent experiments, and represented as mean ± SEM. Statistical analysis was performed by an unpaired t-test for graphs (a–f). (^*p < 0.05, ^**p < 0.009, ^***p < 0.0007).

Discussion

Bacterial invasion of the lung activates the innate immune system, triggering a proinflammatory response and the recruitment of myeloid cells to bolster host immune defense. The initial response must be rapid yet balanced to enable the effective containment and removal of bacteria while maintaining delicate tissue structure and proper lung function, thereby allowing full recovery. S. pneumoniae is the leading cause of bacterial CAP worldwide and is also more prevalent in regions of Zn deficiency^42,43. Inadequate dietary intake in humans increases susceptibility to bacterial infection¹⁰, and Zn supplementation reduces the incidence and impact of infection-based disease⁴⁴; however, the molecular basis for improved outcomes remains unclear. Recently, it was reported that dietary Zn restriction adversely impacts pulmonary Zn biodistribution post-infection, thereby enhancing S. pneumoniae virulence and the extent of infection⁴⁵. The activation and infiltration of murine phagocytic cells were significantly altered by Zn restriction, as well as their ability to eradicate bacteria from the lung.

ZIPs and ZnTs are the primary regulators of Zn biodistribution in mammals and collectively assist in regulating the activity of enzymes, receptors, transcription factors, and signaling pathways via Zn mobilization in and out of the cytosol and organelles^14,15. Among all ZIPs, ZIP8 is unique in that it is the only ZIP whose expression is significantly induced upon pathogen detection by myeloid and parenchymal cells and is required for effective host defense against pathogens^16–19. More recently, our group revealed that pneumococcal pneumonia resulted in an exaggerated inflammatory response in the lung of myeloid-specific Zip8KO mice, including increased recruitment of myeloid cells. Despite increased macrophage numbers, lung bacterial burden increased, which was associated with increased lung tissue damage and dissemination to distal organs²³. This investigation focused on those observations to determine how deficits in Zn homeostasis result in increased bacterial burden despite increased macrophage numbers in the lung. Using single-cell sequence analysis of lung tissue, we reveal that a unique population of dysfunctional proinflammatory macrophages (cluster 10) emerges only in the lung of Zip8KO mice following infection. Inspection of this population predicted that these lung macrophages possess deficits in phagosome formation that we postulated were due in part to decreased intracellular Zn. Indeed, others have shown that Zn deficiency adversely affects phagocytic function in macrophages^46,47 and that Zn supplementation restores bacterial clearance^48–50. Consistent with past studies, we observed that macrophage lysosomal biogenesis and function are impaired due to loss of ZIP8, resulting in reduced lysosome numbers, decreased enzymatic activity, and increased intracellular bacterial accumulation soon after infection. Using inductively coupled plasma mass spectrometry (ICP-MS) analysis, we further demonstrate that Zip8KO macrophages do not increase intracellular pools of Zn upon infection, which is essential for generating a balanced and effective innate immune response^17,19. This effect was also pharmacologically reproduced following intracellular Zn chelation in WT macrophages prior to infection. Although this provides strong evidence that ZIP8 loss causes Zn-dependent dysfunctional lysosome biogenesis, we cannot rule out that other divalent cations may contribute since ZIP8 has previously been shown to effectively transport manganese and iron^51–53. In fact, our studies reveal that iron and manganese levels rapidly elevate in infected WT, but not Zip8KO macrophages (Supplementary Data 1 and Fig. 4a, b), indicating that a lack of intracellular accumulation of these metals may also contribute to deficits in bacterial clearance although, unlike Zn, Fe and Mn levels were the same as infected WT BMDMs at later time points.

Lysosome biogenesis is one of the most important mechanisms for adaptation to pathogen invasion and host defense^32,54. Increased lysosome numbers and enzymatic removal of pathogens accommodate cellular requirements to bolster host defense³⁰. To meet host demand, lysosomes increase their numbers and size by transcriptional activation of lysosomal genes. This is mainly achieved by coordinated actions of transcription factors, transcription repressors, and epigenetic regulators. TFEB plays a central role in the regulation of lysosomal biogenesis. Subcellular localization and activity of TFEB are regulated by kinases that include mTORC1, which resides on the lysosomal surface^55,56. Under normal conditions, when active, mTORC1 phosphorylates TFEB, creating docking sites for the 14-3-3 scaffolding protein, which subsequently retains TFEB in the cytosol³³. Upon upstream signals, including infection³², TFEB is dephosphorylated and released, allowing it to translocate to the nucleus to activate the expression of numerous lysosome-associated genes, as well as genes key to innate immune defense³³. The promoter region of genes that control lysosomal biogenesis harbors common 10-base E-box-like palindromic sequences called coordinated lysosomal expression and regulation (CLEAR) elements, to which TFEB binds, activating gene transcription²⁸. In this study, we report that nuclear localization of TFEB in infected Zip8KO BMDMs is significantly decreased compared to infected WT BMDMs, which also corresponds with a significant decrease in lysosomal enzymatic activity. This is at least in part due to mTORC1 since inhibition of mTORC1 with rapamycin restored the ability of Zip8KO BMDMs to reduce intracellular bacterial burden. Coupled with defects in lysosome numbers and enzymatic activity, these findings reveal for the first time that ZIP8 is required for proper TFEB-mediated activation of lysosomal biogenesis and the successful eradication of bacteria following phagocytic uptake by macrophages. However, TFEB is regulated by other pre-transcriptional, post-transcriptional, and post-translational factors (as recently reviewed⁵⁷), so we cannot exclude the possibility that Zn alone under the conditions studied may also involve non-transcription-dependent events that influence lysosomal activity. Others have shown that Zn regulates V-ATPase assembly on lysosomes by the Zn transporter ZnT2 through coordination of lysosome formation and acidification⁵⁸, and that ZnT3 becomes part of an HC/KC-ATPase/ZnT3 complex that is recruited to lysosomes, and this complex serves as a substitute proton pump for lysosomes⁵⁹. Related to our findings, it has also been reported that Zn increases the activity of cathepsin B and D via lysosomal acidification⁶⁰. Collectively, these findings, in addition to ours, suggest that Zn likely contributes in both gene expression-dependent and -independent ways to influence a coordinated and effective host response against bacterial invasion.

Zn dyshomeostasis adversely impacts the composition of the gut microbiome, leading to decreased production of the SCFA butyrate^35,37. Little is known about how host genetic factors affect SCFA metabolism; however, there are many examples where altered dietary intake and particularly fiber consumption impact microbial communities and the ability of bacteria to manufacture BA and other SCFAs⁶¹. This is important because SCFAs regulate mucosal immunity by impacting immune and non-immune cell targets. BA plays an integral role in maintaining gut epithelial barrier function at mucosal surfaces. BA also influences the antimicrobial and inflammatory function of monocytes and macrophages. Similarly, decreased BA production in the intestine is directly associated with susceptibility to inflammatory and allergic-based gastrointestinal diseases⁶². Of all SCFAs, BA is the most studied, possessing multiple immunoregulatory beneficial effects across multiple species, including mice and humans^63,64. In particular, BA signals through surface-expressed free fatty acid receptors and G protein-coupled receptors, which are expressed by many cells, including macrophages⁶⁵. BA also increases histone acetylation activity, thereby increasing gene expression^36,66,67. Collectively, BA has a broad impact given its ability to influence cell cycle, differentiation, anti-bacterial responses, inflammation, and oxidative stress on immune and non-immune cells^36,68,69. Whereas substantial evidence supports the benefits of BA on the gut, investigation of the gut-lung axis is much less studied, and to our understanding, this is the first study that examines this in the context of Zn dyshomeostasis. In this investigation, we reveal that at baseline, the metabolic profile, and particularly the butanoate synthesis pathway, is altered in lung macrophages in Zip8KO mice. This observation is consistent with previously published work revealing that the gut microbiome of these mice is significantly altered and impaired in its ability to manufacture BA³⁷. Accordingly, this would suggest that systemic BA levels are decreased, adversely impacting mucosal immunity in the lung. Consistent with this, we observed a significant increase in lung bacterial burden, as well as apoptotic macrophages and parenchymal cells in the lung of infected Zip8KO mice, which also corresponded with increased tissue damage and elevated circulating levels of SP-D. Most striking, PBA supplementation for one month reduced bacterial burden, lung tissue cell death, and improved lung mucosal integrity. Based on our modified clinical assessment panel, as early as 24 h postinfection, PBA recipient Zip8KO mice were symptomatically much improved compared to infected Zip8KO mice that did not receive PBA treatment. We did not observe any benefit of PBA supplementation when shorter dosage regimens were evaluated. Similarly, restoration of the ability of Zip8KO BMDMs to clear bacteria required PBA supplementation for 7 days during differentiation and was ineffective when administered after differentiation, before infection in culture. A likely explanation of these findings was first reported by Shulthess and colleagues, who revealed that BA imprints potent antimicrobial properties during gut macrophage differentiation through modulation of histone deacetylation activity³⁶. Taken together, this suggests that dietary- or genetically-induced Zn deficiency promotes dysbiosis resulting in deficient butyrate production by bacteria in the gut, leading to a systemic BA deficit that impairs mucosal immunity in the lung. Based on our findings, reversal of ZIP8-/Zn-mediated defects in lung macrophages, and possibly other myeloid cells, requires prolonged BA exposure to systemically and/or locally reprogram progenitor cells before differentiation into resident macrophages. In turn, cells are better equipped to remove bacteria, balance the proinflammatory response, and minimize extensive tissue damage in the lung. This would be consistent with a recent study revealing that BA suppressed the activation of both murine- and human-derived alveolar macrophages via histone deacetylase inhibition⁷⁰ and other studies demonstrating that antibiotic use, which depletes butyrate-producing bacteria in the gut, increases susceptibility to asthma^71–73. Consistent with our findings, both propionate and butyrate were shown to impact myeloid precursors that bolstered lung defense against influenza infection in mice. Similar to our results, boosting SCFA levels in the intestine enhanced the generation of Ly6C—monocytes in the bone marrow, thereby positively impacting lung macrophage differentiation and function, resulting in reduced neutrophil recruitment to the airways and less tissue pathology⁷⁴. The SCFA acetate has also been shown to reduce the incidence of secondary bacterial infections with influenza A and S. pneumoniae in the lung by enhancing the bactericidal activity, thereby reducing bacterial loads and minimizing lung pathology⁷⁵.

The ZIP8 variant rs13107325 occurs at an allelic frequency of approximately 5% in the American population and 10% in Northern European populations^76,77. It is a nonsynonymous SNP located on exon 8, with the major allele having a cytosine (C) and the minor allele a thymidine (T) at the first position of the codon for residue 391. It is one of the top 10 pleiotropic single-nucleotide polymorphisms (SNPs) associated with inflammation-based diseases in humans²¹. One GWAS investigation revealed that it is strongly linked to an increased risk of Staphylococcus aureus infection²². It has been proposed that rs13107325 alters protein function, although the precise molecular basis for its association with disease traits remains to be determined⁷⁸. Given the relatively high frequency of both the rs13107325 allele and dietary zinc deficiency, more investigation is necessary to ascertain the impact of cellular Zn deficits on the gut-lung axis and its impact on mucosal and innate immune function. Using BA as a leading example from the microbial side and the corresponding alteration of the butanoate synthesis pathway on the host side, we are in the process of identifying and evaluating additional microbial- and host-derived metabolic factors for their potential to restore proper immune function in the setting of Zn deficiency. Whether patients who harbor the rs13107325 allele with or without insufficient dietary zinc intake are more vulnerable to CAP with worse outcomes remains an overarching goal for future studies. In turn, we contend that this will help determine whether aggressive zinc supplementation or, as we show here, alternative treatment strategies can prevent or reduce the incidence of CAP in the commonly occurring setting of Zn deficiency.

Study limitations

Here, we focused on the role of ZIP8 in macrophages since our previous studies in a well-characterized Zip8KO mouse model revealed increased bacterial burden in the lung despite a significant increase in macrophages. In this investigation, we utilized multiple quality control measures to first validate lung tissue and whole-body phenotype before proceeding with scRNAseq analysis to broaden our capacity to identify mechanisms that account for this phenomenon. Although the sample size per treatment condition was limited to one representative specimen, it did reveal predictions of potential mechanisms for macrophage-mediated defects in bacterial clearance that were then validated by in vitro studies in BMDM cultures. In addition, we leveraged the scRNAseq-lung database using a GSM modeling framework. The rationale is that GSMs are explicitly designed to integrate transcriptomic data to infer feasible steady-state reaction activities and metabolite pool changes. While transcript levels do not deterministically specify flux, transcriptionally constrained GSMs reliably recover biologically meaningful metabolic alterations and have been successfully applied many times to reveal mechanisms and therapeutic opportunities across multiple systems. Indeed, one of the key points of this investigation is that transcriptomics alone generated a metabolite-level prediction that reduced butyrate availability is sensed by lung macrophages. This is consistent with our previous studies showing decreased butyrate production by the gut microbiota of Zip8KO mice, and in this investigation, it was further validated by showing the therapeutic effect of PBA supplementation in Zip8KO mice. This concordance between model predictions and empirical results provides strong evidence that the inferred metabolic shifts are not spurious and that the GSM captured essential features of the Zip8KO metabolic phenotype. However, we recognize that GSM models have inherent limitations. Most notably, they cannot capture post-translational modifications, dynamic regulatory events, or interactions with signaling pathways. In addition, GSM models often rely on steady-state assumptions and curated reaction networks, which may not fully reflect context-specific metabolic states or transient cellular responses. They also do not account for protein abundance, enzyme kinetics, compartmental regulation, or spatial organization within the cell, all of which can influence metabolic outcomes. In future studies, when focus is placed on ZIP8-mediated regulation of cellular metabolism, integration of metabolomics with proteomics will be required to allow refinement and validation.

Methods

Animals and care

At the University of Nebraska Medical Center’s (UNMC) Animal Resource Facility, every animal was kept in pathogen-free environments. Water and food were available at all times. The Institutional Animal Care and Use Committee approved the study protocol employed in these studies. The Office of Laboratory Animal Welfare and the National Institutes of Health provided recommendations for all procedures and methods involving animal care used in this study protocol.

Conditional Zip8 knockout mice, or Zip8KO for short, were produced using the methods previously mentioned⁷⁹. In short, the Neo cassette next to the upstream loxP site was deleted by breeding heterozygous Zip8^flox-neo/+ mice to ROSA26:FLPe knock-in mice (The Jackson Laboratory) that expressed FLP1 recombinase ubiquitously. Zip8^flox/flox mice were created by mating the resulting Zip8^flox/+ mice. PCR and DNA sequencing validated the loxP sites flanking exon 3 and confirmed the removal of the FRT-flanked region. To create the conditional Zip8KO, Zip8^flox/flox mice were bred to LysMcre (The Jackson Laboratory), which is specific to myeloid cells. Jackson Labs provided C57BL/6J wild-type equivalents, which were produced specifically for research purposes.

WT and Zip8KO mice received either saline or 30 mM sodium-4-phenylbutyrate (PBA, cat# 11323, Caymen Chemical) in their drinking water for 4 weeks. The water was changed every 7 days. The PBS or PBA-treated mice were instilled with 4 × 10⁸ CFU of S. pneumonia as described below. The PBA-treated uninfected groups and the infected groups at 24 h post-infection received a single bolus intraperitoneal (IP) injection of PBA at a dose of 400 mg/kg. Following the bolus dose, the mice were euthanized at 48 h post-infection. Control groups received an equal volume of PBS via intraperitoneal injection.

Culture, quantification, and Instillation of S. pneumoniae

S. pneumoniae strain JWV500 (D39hlpA-gfp-Cam’), a generous gift from Dr. Jan-Willem Veening (University of Lausanne, Switzerland), was grown to mid-log phase, aliquoted, frozen, and stored at −80 °C until further use. For lung infection studies, bacteria were grown to log phase in Remel Mueller Hinton Broth (cat# R112478, Fisher Scientific, Lenexa, KS) supplemented with 32 mg/mL chloramphenicol to an OD₅₄₀ of 1.0. For quantification of pneumococci, serial dilutions of the bacteria were plated on Remel blood agar plates (cat# R01202, ThermoFisher), incubated at 37 °C with 5% CO₂ overnight to determine colony-forming units (CFUs). For intratracheal instillations, mice were lightly anesthetized using 2% isoflurane and 1 L/min of oxygen and instilled with 4 × 10⁸ CFU of S. pneumoniae in 100 µL of PBS. The inoculum was confirmed by serial dilutions.

Single-cell RNA sequencing

Lungs from 1 mouse per treatment group were processed as previously described in ref. ⁸⁰. Briefly, the lungs were perfused with 10 mL of heparin-PBS. The GentleMACS dissociator (Miltenyi Biotech) was used to separate the harvested lungs from the digesting solution (collagenase I, 0.2 μg/μL + DNase I, 75 U/mL + heparin, 1.5 U/mL) in Dulbecco’s Modified Eagle’s Media (DMEM). Samples were shaken for 30 min at 37 °C. PBS with 4 mM EDTA was used to counteract the activity of the digestion solution. Red blood cells were lysed for 1 min in 1 mL of lysis solution (BioLegend), and then neutralized with ice-cold Gibco DMEM. Dead cells were removed using the Dead Cell Removal Kit (Miltenyi Biotec). FACS buffer (2% fetal bovine serum (FBS) + 0.1% NaN₃ in PBS) was used to prepare cells for RNAseq. All reagents, unless specified, were purchased from Sigma. The viability and quantification of single-cell suspensions obtained from the whole lung were determined using a LUNA-FLTM Dual Fluorescence Cell Counter (Logos Biosystems). Following manufacturer recommendations, single cells were extracted from cell suspensions (100–2000 cells/μL) using a 10× Chromium controller (10× Genomics). The gel beads emulsion (GEM)/sample solution was collected and placed into strip tubes. Following cDNA amplification, reverse transcription was carried out using a thermocycler (C1000 TouchTM Thermal Cycler, Bio-Rad). Amplified products were solid phase reversible immobilization (SPRI) bead-purified and evaluated by a Fragment Analyzer (Agilent, Santa Clara, CA). Twenty-five percent of the cDNA volume was fragmented, and PCR purification and clean-up were carried out using double-sided SPRIselect (Beckman Coulter). Following adapter ligation, each sample underwent SPRI clean-up and PCR amplification utilizing sample-specific indexes. The Fragment Analyzer was used to quantify, purify, and determine the library size distribution of PCR products. PCR products were purified, quantified, and sequenced on an Illumina (San Diego, CA, USA) Novaseq6000. Libraries were sequenced to an average depth of 50,000 reads per cell following manufacturer-recommended conditions. To assess macrophage polarization states (proinflammatory M1-like and anti-inflammatory M2-like) across the 24 identified clusters, we defined gene sets representing proinflammatory (M1-like)(CD80, CD86, CD64, CD32, iNOS, CD40, CD83, MHCII, TLR4, IL1-R1, SOC3, TLR2) and anti-inflammatory (M2-like)(CD163, CD206, Arg1, Fizz1, Ym1, CD115, PPARg, CD301, PDL2) states. For each cluster, the transcriptomic score was calculated as the mean expression of the selected marker genes within that cluster. The expression value for each gene was the normalized unique molecular identifier (UMI) count, adjusted for cluster size.

Bioinformatic analysis

The 10× Genomics cellranger (version 7.1) analysis pipeline was used for demultiplexing and generating a gene-barcode matrix. Sub-function mkfastq from cellranger was used to generate raw read files in FASTQ format; sub-function count was used to map the raw reads to the mouse reference genome (version 10), and then performed filtering, cell barcode counting, and UMI counting. As a result, a gene expression matrix was generated containing the raw UMI count for each gene for each cell per sample. After generating matrices for all samples, sub-function aggregation was called to aggregate all four samples. All matrices were normalized by subsampling reads from higher-depth GEM wells to make them comparable. The aggregated matrix is then fed into the 10° Loupe Browser for downstream analysis. Cells with more than 5% mitochondrial reads are filtered out. Then applied principal component analysis (PCA) was applied to reduce the dimensions, and the first 10 PCs are used to generate the t-SNE plot and perform the graph-based clustering. The IPA was performed on the gene lists generated from selected clusters. The lists are filtered based on the average expression value (>1) and log fold change > 0.58 and <−0.58.

Clinical assessment score

We adapted the modified mouse clinical assessment score (MMCAS) for pneumococcal pneumonia from the modified murine sepsis score and modified MMCAS as described by Mai et al.⁸¹. The MMCAS generates an objective score from an average of 6 components described in Supplementary Table 4. We standardized the MMCAS component scores to a five-point scale ranging from 0 to 4. Each mouse was scored in a blinded fashion by the pathologist.

BALF analyses

Lungs were lavaged three times with 1 mL of ice-cold PBS. Total BAL cell counts were obtained using a TC20 automated cell counter (Bio-Rad). Differential cell counts were determined using Cytospin-prepared slides stained with Hema-3 (cat# 22122911, Thermo Fisher Scientific). Cytokine and chemokine levels were measured using commercially available ELISA kits according to manufacturer instructions (BioLegend, R&D Systems).

Tissue immunostaining

Formalin-fixed paraffin-embedded lung sections were stained for in situ apoptosis using the Click-IT Plus TUNEL Assay Alexa 594 (cat# C10618, ThermoFisher) according to manufacturer instructions. Lung sections were co-stained for the macrophage marker F4/80 (cat# 14480182, ThermoFisher) and nuclei with DAPI after being embedded in VectaShield Vibrance Mounting Medium (cat# H180010, VectorLabs). The secondary antibody against F4/80 was anti-rat IgG Alexa 647 (cat# 712605150, VWR), and S. pneumoniae were GFP-tagged. Images were captured using a Leica Thunder Imager 3D Live Cell Microscope (Leica Microsystems) equipped with LASX software and a Hamamatsu Orca Flash 4 camera (Hamamatsu). Imaging was performed in a wide-field using a dry 63× Plan Fluotar 0.7NA objective. Illumination was provided using an LED 5 with an output of ~200 mW at wavelengths of 390, 480, 555, and 630 nm. The light was passed through a quad filter cube DFT51010 (ex 391/479/554 and 638 nm; dichroic of 415/500/572 and 660 nm and emission 435/519/594 and 695 nm) before encountering the sample on the microscope stage. Image processing and analyses were done using ImageJ FIJI software (Schindelin, 2012 #14241).

Flow analytical cell sorting of macrophages in lung tissue

Lung lobes were collected and perfused with digestion solution containing 1x HBSS (Hyclone, GE Healthcare Lifesciences, Logan, UT), 1 mg/mL collagenase D (Roche), and 20 µg/mL DNase (Roche), incubated for 30 min at 37 °C, and homogenized using gentleMACS™ Octo Dissociator (Miltenyi Biotec, Auburn, CA). After enzymatic digestion and red blood cell lysis (1x RBC Lysis Buffer, Invitrogen by Thermo Fisher Scientific, Life Tech Corp, Carlsbad, CA), samples were resuspended in FACS rinsing buffer (1× PBS supplemented with 2% FBS and 0.1% sodium azide). Cell viability was measured using the Zombie UV viability kit (ThermoFisher), according to the manufacturer's protocol. Then cells were stained for cell surface markers. After cell surface staining was completed. Cells were fixed and permeabilized in Cyto-Fast Fix/Perm Buffer Set (BioLegend, cat# 426803) according to the manufacturer's protocol. Intracellular markers were then stained in Cyto-Fast Perm buffer for 20 min at room temperature, washed (1x Perm Buffer, 2x FACS buffer), then resuspended in 0.5 mL FACS buffer and analyzed on BD Fortessa X50. The following antibodies and stains were used for the following markers at the stated dilutions. Ly6G (BV650; 1:400), CD11c (BV711; 1:400), MHCII (BV421; 1:400), Ly6C (BV605; 1:400), CD64 (PE Cy7; 1:200), CD4 (APC Cy7; 1:800), CD8 (Per CP Cy5.5; 1:400), F480 (PE; 1:400), CD19 (BV785; 1:400), CD206 (AF700; 1:400), Zombie Viability (UV; 1:500), and TruStain (15 µL) all from BioLegend; CD45 (BUV805; 1:200), CD3 (BUV615; 1:200), iNOS (BUV737; 1:400), and ARG1 (APC; 1:400) – all from ThermoFisher; CD11b (BV480; 1:400), CD103 (PE CF594; 1:400), and NK1.1 (BUV395; 1:200) – all from BD Biosciences.

Isolation and generation of BMDMs

BMDMs were generated as previously described⁸² with a few modifications. Briefly, the femurs and tibias from C57BL/6 WT and Zip8KO mice (7-10 weeks old) were harvested, and bone marrow cells were plated in DMEM (cat# 10-014-CM, Corning) supplemented with 10% FBS (cat# 35011CV, Corning), 1% Pen/Strep (cat# 16777164, VWR), and 50 ng/mL recombinant mouse granulocyte monocyte-colony stimulating factor (GM-CSF cat# AF-315-03, PeproTech) at a density of 1.4 × 106 cells/mL in 100 mm dishes. In another set of experiments, WT and Zip8KO-derived bone marrow cells were differentiated under the same conditions in the presence of 100 µM sodium-4-PBA for 7 days. On day 7, adherent BMDMs were collected and plated in 6-well cell culture plates for subsequent studies. PBA-differentiated BMDMs were maintained in 100 µM PBA throughout the study. In a separate study, BMDMs were treated with 500 µM or 1000 µM BA (cat# 13121, Caymen Chemical) for 3 h post differentiation.

In parallel studies, BMDMs were also plated at a density of 1 × 10⁶ cells/mL in 6-well plates, followed by 24 h of pre-treatment with vehicle (DMSO), rapamycin (cat#13346, Cayman Chemical; 250 nM), or 30 min pre-treatment with the intracellular Zn chelating agent Tris 2-pyridylmethyl amine (TPA, cat#723134, Sigma).

Assessment of bacterial burden

BMDMs were plated at a density of 1 × 10⁶ cells/mL in 6-well plates and allowed to adhere for 24 h. Cells were then infected with S. pneumoniae (multiplicity of infection of 10) for 15, 30, and 60 min. Cells were then washed with PBS and treated with 200 µg/mL gentamicin and 10 µg/mL penicillin/streptomycin for 30 min to kill the bound and extracellular bacteria. Cells were washed with PBS and lysed with water. Serial dilutions of lysates were plated on blood agar plates, and following overnight incubation, colony-forming units (CFUs) were enumerated.

In separate experiments, BMDMs were plated at a density of 5 × 10⁵ cells/mL on poly-lysine-coated coverslips in 6-well plates. S. pneumoniae were labeled with 20 µM CFSE (cat# 423801, BioLegend) at 37 °C for 30 min. CFSE-labeled bacteria were used to infect BMDMs (multiplicity of infection of 10) for 15, 30, and 60 min. Cells were then washed with PBS three times and stained with LysoTracker^TM Red DND-99 (cat# L7528, ThermoFisher) for 1 h, followed by washing with PBS, fixation in 4% paraformaldehyde. Coverslips were then mounted on microscope slides, and nuclei were stained with DAPI-containing mounting media (cat# H180010, VectorLabs). Images were visualized at 40× (oil immersion lens) using a Zeiss Observer Z1 inverted phase contrast fluorescence microscope (Carl Zeiss Microscopy), and fluorescent intensity was quantified using ImageJ FIJI software (National Institutes of Health).

Western blotting

Cell lysates were obtained using RIPA buffer (cat# 9806, Cell Signaling) containing protease and phosphatase inhibitors. Cytoplasmic and nuclear proteins were isolated using a commercial kit (ThermoFisher #78835) with Halt protease/phosphatase inhibitors (ThermoFisher #1861281). Proteins were transferred to nitrocellulose membranes (Bio-Rad) following SDS-PAGE separation and blocked with Li-Cor TBS blocking buffer (Li-Cor # 927-60001). After blocking, membranes were probed with primary antibodies overnight at 4˚C. The following primary antibodies were used: pmTOR (cat#5536S, Cell Signaling, 1:1000), mTOR (cat#2983S, Cell Signaling, 1:1000), pTFEB (cat#87932S, Cell Signaling, 1:1000), TFEB (cat#83010S, Cell Signaling, 1:1000), cathepsin B (Cat# 31718S, Cell Signaling, 1:1000), cathepsin L (cat#sc-390367, Santa Cruz, 1:1000), β-actin (Cell Signaling, cat#8457S, 1:5000) GAPDH (ThermoFisher #MA515738, 1:2000), and LaminB1 (BioLegend #869802, 1:2000). Following TBST washing (x3) membranes were then incubated for 1 h at room temperature with an anti-rabbit horseradish peroxidase-linked antibody (cat# 7074, Cell Signaling, 1:5000) or anti-mouse horseradish peroxidase-linked antibody (cat# 7076, Cell Signaling, 1:5000). Proteins were treated with WesternSure Premium (Li-Cor #926-95000) chemiluminescence substrate and exposed to a luminescent image analyzer (Li-Cor C-Digit) to visualize and quantify proteins.

Zinc quantification

Intracellular Zn was quantified at baseline and in infected WT and Zip8KO BMDMs at 0 and 6 h post-infection with S. pneumonia (multiplicity of infection of 10) in collaboration with the UNMC Nanomaterials Characterization Core via inductively coupled plasma mass spectrometry (ICP-MS). Briefly, samples were digested using a 5:1 mixture of nitric acid (OPTIMA grade, 70%, Fisher Scientific) and ultrapure hydrogen peroxide (ULTREX II, 30%, ThermoFisher). Samples were then placed into 2 mL boil-proof polypropylene tubes and digested overnight. All samples were heated to 84^oC until dry. Samples were then resuspended in a matrix (2% nitric acid, containing an internal standard), and placed in a sonication bath at room temperature for 30 min. Samples were centrifuged, and supernatants were collected for quantification of intracellular Zn content. Calibration standards and samples were prepared in an acid matrix of 2% OPTIMA grade nitric acid. Calibration standards for Zn were prepared using single-element calibration standards (SCP Science) to obtain a six-point calibration curve. Rhodium at a final concentration of 10 ppb, was added to standards, blanks, and samples and used as an internal standard to correct for potential sample matrix and/or nebulization effects.

Zinc measurement by atomic absorption spectroscopy

Tissue Samples were dried, weighed, and then digested in 1 mL Nitric Acid: Perchloric Acid (1:2) for 2 h at 80 °C. Samples were diluted with MilliQ, then subjected to atomic absorption spectroscopy (AAS) using the PinAAcle 500 Atomic Absorption Spectrometer (Perkin Elmer).

LAMP1 immunofluorescence and quantification

BMDMs were seeded onto 12 mm poly-L-lysine-coated glass coverslips in 12-well plates and allowed to adhere overnight. Cells were washed with PBS, then infected with S. pneumoniae at an MO1 of 10:1, then incubated at 37 °C, 5% CO₂ for designated time points. Cells were washed in PBS, fixed in 4% PFA, followed by permeabilization and block (BupH PBS, 0.1% saponin, 1% NMS, 4% DIG block, and 1% BSA). Primary antibody was incubated overnight at 4 °C (LAMP1 clone 1D4B, Developmental Studies Hybridoma Bank, University of Iowa), followed by secondary staining using Alexa IgG donkey anti-rat 594 (A21209, ThermoFisher). Coverslips were washed in PBS and mounted onto slides using mounting medium containing DAPI (VectaShield). Images were visualized at 40× oil using a Zeiss Observer Z1 inverted phase contrast fluorescence microscope (Carl Zeiss Microscopy). QUPATH version 6.0, an open-source platform for bioimaging analysis, was used for quantification. Projects were created by importing CZI files. Cell detection was performed by the following script: full figure annotation> cell detection (LAMP1 Alexa 594, pixel size 0.5 microns, sigma 1.5, min-max area 10-400, threshold 100 without split by shape, & cell expansion 5 microns). Objects without a nucleus were excluded from analysis.

DQ red BSA assay

BMDMs were infected with S. pneumoniae (multiplicity of infection of 10) for 60 min, washed with PBS, and treated with gentamicin/penicillin/streptomycin as described previously. Cells were then washed with PBS and incubated with 10 µg/mL DQ Red BSA dye (cat#D12051, ThermoFisher Scientific) for 18 h, after which cells were fixed with 4% paraformaldehyde and nuclei stained with DAPI containing mounting media (VectaShield). Images were visualized at 40× (oil immersion lens) using a Zeiss Observer Z1 inverted phase contrast fluorescence microscope (Carl Zeiss Microscopy), and fluorescent intensity was quantified using ImageJ FIJI software (National Institutes of Health).

Cathepsin B and L activity assays

BMDMs were infected with S. pneumoniae (multiplicity of infection of 10) for 60 min, washed with PBS, and treated with gentamicin/penicillin/streptomycin as previously described. Cells were then washed with PBS and incubated in DMEM and 10%FBS for 18 h. Lysates were analyzed for cathepsin B (cat#ab65300, Abcam) and L (cat#MAK401, Sigma) activity according to manufacturer instructions.

ELISA

BALF cytokines (IL-6 Cat# 431304, IL-1β cat# 432604, and CXCL1 cat# DY453-05) and serum SP-D (cat# MSFPD0) concentrations were determined using commercially available ELISA kits (BioLegend and R&D Systems) according to manufacturer instructions.

TFEB translocation (flow cytometry)

WT and Zip8KO BMDMs were generated as previously described and differentiated with and without 100 µM PBA. Cells were plated at a density of 1 × 10⁶ cells/mL in 6-well plates, followed by infection with S. pneumoniae (multiplicity of infection of 10) for 60 min. Cells were washed and immediately fixed and washed in ice-cold Cyto-Fast Fix/Perm Buffer (PB) set per manufacturer protocol (cat# 426803, BioLegend). Following fixation and wash, cells were resuspended in PB buffer (MACS Rinsing Solution cat# 130-091-022 &, MACS BSA solution cat# 130-091-376, Miltenyi Biotec) and stained for TFEB Alexa 647 (cat# ab2832744, AbCam) for 30 min in the dark, then washed. Cells were resuspended in PB staining buffer, and nuclei were stained with DAPI (cat# sc-3598, Santa Cruz Biotechnologies). Samples were then analyzed using an Amnis FlowSight Imaging Flow Cytometer (Luminex). Five thousand cells were enumerated per sample in each experiment and then shown as the percentage total TFEB-nuclear positive cells.

Generation of GSM models

We used the single-cell RNA sequencing datasets of all the conditions (pre- and post-infection wildtype and knockout mice) to generate macrophage cluster (1, 3, 4, 5, and 10) specific GSM prediction models by using the ftINIT method⁸³. The ftINIT method efficiently generates context-specific GSM models by integrating multi-omics data while preserving essential metabolic functionality and ensuring biological relevance. It balances computational efficiency with flexibility, making it ideal for studying condition-specific metabolism, disease modeling, and therapeutic target identification. This approach produces robust, scalable models tailored to specific biological conditions or environments. Hence, we used the reconstructed models to predict the metabolic landscape of the macrophage-specific clusters for the WT and Zip8KO models’ pre-infection and post-infection.

Flux balance, flux variability, and FSA

FBA³⁸, FVA³⁹, and FSA⁴⁰ were performed to estimate the flux ranges for reactions and compute the pool sizes of the metabolites. FBA was used to determine the optimal flux distribution under specific conditions, FVA provided the range of fluxes for reactions that satisfy the optimal solution, and FSA calculated the metabolite pool sizes at different conditions (WT and Zip8KO pre and post-infection).

Statistical analyses

Data were analyzed using GraphPad Prism version 10.2.3 (GraphPad Software, La Jolla, CA). An unpaired Student’s t-test was used to determine differences between groups. Multiple group comparisons were made using ANOVA with Tukey’s post-hoc tests. Values are expressed as means ± SEM. A value of p < 0.05 was considered significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Conceptualization: D.L.K. and D.R.S. (Derrick R. Samuelson). Methodology: D.R.S. (Deandra R. Smith), S.H. (Sabah Haq), D.R.S., M.N., S.M., R.S., A.P., P.B., S.R., D.L.K. Validation: D.R.S. (Deandra R. Smith), M.N., S.M., R.S., and D.L.K. Formal analysis: D.R.S. (Deandra R. Smith), D.R.S. (Derrick R. Samuelson), M.N., S.M., R.S., and D.L.K. Resources: D.R.S. (Derrick R. Samuelson), D.L.K. Data curation: D.R.S. (Deandra R. Smith), D.R.S. (Derrick R. Samuelson), and D.L.K. Writing—original draft preparation: D.R.S. (Deandra R. Smith) and D.L.K. Writing—review and editing: D.R.S. (Deandra R. Smith), D.R.S. (Derrick R. Samuelson), and D.L.K. Visualization: D.R.S. (Deandra R. Smith), D.R.S. (Derrick R. Samuelson), M.N., S.M., R.S., A.P., P.B., and D.L.K. Supervision: D.L.K. Project administration: D.L.K. Funding acquisition: D.R.S. (Derrick R. Samuelson) and D.L.K. All authors have read and agreed to the published version of the manuscript.

Peer review

Peer review information

Communications Biology thanks Ajitha Thanabalasuriar and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Jesmond Dalli and Mengtan Xing. A peer review file is available.

Data availability

FASTQ files and processed sequencing data have been deposited in the National Center for Biotechnology Information Sequence Read Archive (GSE313588). Additional data that supports the findings is in the source Data Excel file.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-025-09504-8.